ray.rllib.policy.sample_batch.MultiAgentBatch.agent_steps#

MultiAgentBatch.agent_steps() → int[source]#

The number of agent steps (there are >= 1 agent steps per env step).

Returns

The number of agent steps total in this batch.